The fact that you say that your heatsink temp with no signal is up at 42 degrees sujests that you might have a bit of instability. I find that mine run at room temp unless driven reasonably hard (though I have very sensitive speakers). If there is instability it might go a way to explain the harsh top end - as this is likely to be the part that suffers most from instability.
What type of box is it in ? Is it adequately shielded ?
You might try bypassing on the chip with ceramics.
Good grounding is absolutely essential.
Keep us posted.
Shoog
What type of box is it in ? Is it adequately shielded ?
You might try bypassing on the chip with ceramics.
Good grounding is absolutely essential.
Keep us posted.
Shoog
Shoog said:
I think you're right.
Ashok,
Why not also decrease the 10ohm res. of the Zobel to 2.7 ohm? There must be some kind of instability/oscillation happening or RF noise getting in.
What kind of Spkr cables are you using? May be a high capacitance one.
Is it possible to see the exact sch. with the volume ctrl?
/Greg
More on my IGC amp.
Greg,
I had a look on the scope with a speaker load and 4 cycle sinewave burst at various frequencies and from low level to full level including some clipped signals. No sign of any RF oscillation. The scope goes to 100Mhz. However I did find that some of the roughness at times was due to a loose bass unit on the speaker. I put in a new gasket and remounted the driver. That problem went away.
The sound is now much better but still does not sound 'nice' to put it simply. I have been using the chip amp by itself as an IGC. Zin is about 56K with 0.22uF cap for dc decoupling. It's a PIO cap. At moderate listening levels it does sound nice as long as the music is not too complex.
I will now change the CD player , add some more decoupling ( more film caps on the board ) .
Shoog,
The grounding is fine. The unit is open and the pcb sits on top of an Aluminum sheet ( grounded) with an insulator in between (new FRP board). The scope shows no RF on the test signals at any level .
But the heat sink does get quite hot even now. I have used a mica insulator between the chip and the heat sink with silicone grease on both sides.
There is nothing to the schematic.
Cin is 0.22uF
R in is 56K ( to inv input.)
R feedback is 100k ( -ve input to output voltage divider of 2.2K and 75 ohms) That is a T network feedback.
Positive input to ground via 100K .
470uF on each supply line with 0.1uF film on the pcb side on each 470uF. Mute is 22K with 100uF cap.
47K log volume is connected to the 0.22uF input cap.
Response seems to be -0.4db at 20Hz and 20KHz with speaker load. Output impedance must be quite low. Didn't measure it. The response with speaker load is quite flat and so it must be quite low.
Cheers.
Ashok.
Greg,
I had a look on the scope with a speaker load and 4 cycle sinewave burst at various frequencies and from low level to full level including some clipped signals. No sign of any RF oscillation. The scope goes to 100Mhz. However I did find that some of the roughness at times was due to a loose bass unit on the speaker. I put in a new gasket and remounted the driver. That problem went away.
The sound is now much better but still does not sound 'nice' to put it simply. I have been using the chip amp by itself as an IGC. Zin is about 56K with 0.22uF cap for dc decoupling. It's a PIO cap. At moderate listening levels it does sound nice as long as the music is not too complex.
I will now change the CD player , add some more decoupling ( more film caps on the board ) .
Shoog,
The grounding is fine. The unit is open and the pcb sits on top of an Aluminum sheet ( grounded) with an insulator in between (new FRP board). The scope shows no RF on the test signals at any level .
But the heat sink does get quite hot even now. I have used a mica insulator between the chip and the heat sink with silicone grease on both sides.
There is nothing to the schematic.
Cin is 0.22uF
R in is 56K ( to inv input.)
R feedback is 100k ( -ve input to output voltage divider of 2.2K and 75 ohms) That is a T network feedback.
Positive input to ground via 100K .
470uF on each supply line with 0.1uF film on the pcb side on each 470uF. Mute is 22K with 100uF cap.
47K log volume is connected to the 0.22uF input cap.
Response seems to be -0.4db at 20Hz and 20KHz with speaker load. Output impedance must be quite low. Didn't measure it. The response with speaker load is quite flat and so it must be quite low.
Cheers.
Ashok.
Re: More on my IGC amp.
You can try changing that zobel for something different.
2.7R + 0.1uf from output to ground.
Next step: regulate.😀
I think that the amp is not driving your speakers properly.
ashok said:
You can try changing that zobel for something different.
2.7R + 0.1uf from output to ground.
Next step: regulate.😀
I think that the amp is not driving your speakers properly.
Changed source..............better sound.
Hi Carlos,
I just changed the source. I used an old upgraded Rotel 855 cd player.
It sounds very nice even at higher volume. I think that the hf limiting filter is required at the input .
I could not change the Zobel as I do not have a 2.7 ohm resistor. Maybe a quarter watt - I'll try that till I get a 1 watt 2.7 ohm type. Under normal conditions 1/4 watt should be OK.
It is possible that I'm running the speakers harder to see how it sounds clipped. But the sound did get noisy before clipping started. Not what I get with the older amps. Will have to check all this again.
I think my ears need some rest. I did some listening near the speakers at high volume and my ears feel slightly stuffed . Can't afford to kill them. Will post results of test tomorrow.
Do your heatsinks get very hot ? Shoog felt that it was too hot for normal operation. I must look up my Mission's impedance curve. Maybe that dips to 3 ohms or so. That can cause lots of problems.
My supply voltage is +/- 36 volts. Each channel has two 4700uF caps and a 470uF at the chip. Nuuk suggested that I change all that and have only 1000uF at the chip like most guys are doing. I will have to try that out also. However not just yet.
Thanks to everyone for pitching in to solve the problems.
Cheers,
Ashok.
Hi Carlos,
I just changed the source. I used an old upgraded Rotel 855 cd player.
It sounds very nice even at higher volume. I think that the hf limiting filter is required at the input .
I could not change the Zobel as I do not have a 2.7 ohm resistor. Maybe a quarter watt - I'll try that till I get a 1 watt 2.7 ohm type. Under normal conditions 1/4 watt should be OK.
It is possible that I'm running the speakers harder to see how it sounds clipped. But the sound did get noisy before clipping started. Not what I get with the older amps. Will have to check all this again.
I think my ears need some rest. I did some listening near the speakers at high volume and my ears feel slightly stuffed . Can't afford to kill them. Will post results of test tomorrow.
Do your heatsinks get very hot ? Shoog felt that it was too hot for normal operation. I must look up my Mission's impedance curve. Maybe that dips to 3 ohms or so. That can cause lots of problems.
My supply voltage is +/- 36 volts. Each channel has two 4700uF caps and a 470uF at the chip. Nuuk suggested that I change all that and have only 1000uF at the chip like most guys are doing. I will have to try that out also. However not just yet.
Thanks to everyone for pitching in to solve the problems.
Cheers,
Ashok.
Re: Changed source..............better sound.
Why do I always suppose that people have a good source?
At least use the same source when comparing amps.
That will do fine.
No, not hot.
That's what I meant, the amp is not driving the speakers properly.
That's why you have the heatsinks very
.
Your speakers have impedance dips to 3 ohms and this is a very high voltage.
I'm not surprized that the amp clips like mad at loud volume.
You may be hearing the effect of the SPiKe protection.
You should not pass +/-30~31V.
IMHO this is for most bass-reflex speakers.
ashok said:
Why do I always suppose that people have a good source?
At least use the same source when comparing amps.
ashok said:
That will do fine.
ashok said:
No, not hot.
That's what I meant, the amp is not driving the speakers properly.
ashok said:
That's why you have the heatsinks very
.Your speakers have impedance dips to 3 ohms and this is a very high voltage.
I'm not surprized that the amp clips like mad at loud volume.
You may be hearing the effect of the SPiKe protection.
You should not pass +/-30~31V.
IMHO this is for most bass-reflex speakers.
Re: More on my IGC amp.
According to the values you posted your gain is 53. That's quite a lot. Have you tried to replace the input res (56k) with 100k and that way have more reasonable gain of 30. Even changing the 75 ohms to 100 ohms. When I fiddled with my GC I find the gain of 20 to be better sounding than the gain of 30. Also less gain = less susceptibility to HF noise and more NFB = less distortions.
/Greg
ashok said:
According to the values you posted your gain is 53. That's quite a lot. Have you tried to replace the input res (56k) with 100k and that way have more reasonable gain of 30. Even changing the 75 ohms to 100 ohms. When I fiddled with my GC I find the gain of 20 to be better sounding than the gain of 30. Also less gain = less susceptibility to HF noise and more NFB = less distortions.
/Greg
Are you still there Joe? 😉
I tried the T-network yesterday on a new (LM3875) IGC.
Input 22K, T-network 10K - 10K with 100R to ground, trimmer from NI to ground set to 10K.
Results were 42 mV on one channel and 31 mV on the other.
After adjusting the trimmers and then replacing them with resistors of a similar value, I got 0 mV on one channel and 3.9 mV on the other.
A bit early to comment on overall sound quality with a new amp but clarity is definitely improved.
I tried the T-network yesterday on a new (LM3875) IGC.
Input 22K, T-network 10K - 10K with 100R to ground, trimmer from NI to ground set to 10K.
Results were 42 mV on one channel and 31 mV on the other.
After adjusting the trimmers and then replacing them with resistors of a similar value, I got 0 mV on one channel and 3.9 mV on the other.
A bit early to comment on overall sound quality with a new amp but clarity is definitely improved.
Nuuk said:Are you still there Joe? 😉
I tried the T-network yesterday on a new (LM3875) IGC.
Input 22K, T-network 10K - 10K with 100R to ground, trimmer from NI to ground set to 10K.
... 42 mV on one channel and 31 mV on the other.
After adjusting the trimmers and then replacing them with resistors of a similar value, I got 0 mV on one channel and 3.9 mV on the other.
... clarity is definitely improved.
OK Nick, so you tweaked the value from (+) to ground? How much did they end up deviating from 10K, or putting it differently, what value did you use? I guess you are around 12K or 8k2, correct?
I would still reckon that sticking with 42mV and 31mV is the way to go (assuming the value of the trimmer was at 10K?), as both will be well sub 1mW dissipation in the voice coil and that includes even 3-4 Ohm speakers. In fact with 3 Ohm DC voice coils and 42mV, it's only 0.33mW - insignificant. AC balance is more important, IMHO.
Joe R.
Ohm's law redefined?
Just curious, Joe: How did you arrive at the figure 0.33mW?
My calculations give me 0.588mW for 42mV at 3ohms. Not that big a difference, but still...
Rune
Just curious, Joe: How did you arrive at the figure 0.33mW?
My calculations give me 0.588mW for 42mV at 3ohms. Not that big a difference, but still...
Rune
Spot on for one channel Joe - 8K2. But for the other, 5K6!
I would not be unhappy with 42 mV but I figured if I must have a resistor from NI to ground instead of a piece of wire, then I may as well go for the lowest DC offset at the same time! 😉
Re: Ohm's law redefined?
Uhhh, right you are (add sound of offending calculator thrown into garbage bin).
Hardly worth loosing sleep over it. 🙂
Joe R.
runebivrin said:
Uhhh, right you are (add sound of offending calculator thrown into garbage bin).
Hardly worth loosing sleep over it. 🙂 Joe R.
Some changes...
I added an LPF at the input. It has a 20Khz response of -0.4 db.
The amp sounds very nice now. I also added the tube cathode follower at the input. The cathode follower takes the full input signal. The volume pot comes between the chip and the cathode of the input stage.
We compared it against an Audiolab 8000S integrated amp last night. The chip amp sounds better especially in the midrange and low level detail. Bass on the Audiolab has more transient attack but that is possibly because it does sound a bit brighter also .... and harsh ,in comparison.
Low end on the Audiolab is also a bit deeper. That could be because of the roll off frequency of the chip amp.
I'd pick the chip amp for better overall sound.
Nice , very nice. Only negative point is that it doesn't handle low impedances well at high volume. The protection really sounds awful when it activates. Now on to my TDA7294 chips !
That should come before the new 3886 board.
Cheers.
Ashok.
I added an LPF at the input. It has a 20Khz response of -0.4 db.
The amp sounds very nice now. I also added the tube cathode follower at the input. The cathode follower takes the full input signal. The volume pot comes between the chip and the cathode of the input stage.
We compared it against an Audiolab 8000S integrated amp last night. The chip amp sounds better especially in the midrange and low level detail. Bass on the Audiolab has more transient attack but that is possibly because it does sound a bit brighter also .... and harsh ,in comparison.
Low end on the Audiolab is also a bit deeper. That could be because of the roll off frequency of the chip amp.
I'd pick the chip amp for better overall sound.
Nice , very nice. Only negative point is that it doesn't handle low impedances well at high volume. The protection really sounds awful when it activates. Now on to my TDA7294 chips !
That should come before the new 3886 board.
Cheers.
Ashok.
Re: Some changes...
Probably not the protection circuit, more like clipping. My experience is that the protection causes a reduction in gain, eventually shutting it down if you keep putting volume up.
Re bass, can I say that there is something on the way that will majorly improve the bass, and a whole lot more. I cannot say much more than that yet.
Joe R.
ashok said:
Probably not the protection circuit, more like clipping. My experience is that the protection causes a reduction in gain, eventually shutting it down if you keep putting volume up.
Re bass, can I say that there is something on the way that will majorly improve the bass, and a whole lot more. I cannot say much more than that yet.
Joe R.
By the way guys, you are using inverting mode of amp - studio standard is noninverting phase of all there used equipment. Do you know why ?
Could you check bias again, when the temperature of the chip is different?
Maybe, you compare the psu-caps of the audiolab with your gainclone?
I bet, the Audiolab has higher values, what improves the power bandwidth at low frequencies.
Franz
Hi Joe,
Most of my tests were done under clipping level. I monitored it on a scope. The sound I was getting was like HF trash ( in the audible range). But since I put in the LPF I have not heard that yet. But I did make some other changes also.
Anyway it does sound nice if you don't push it too hard.
I've been toying with the idea of adding external components to improve on what is already there . Let me see how the 7294 fares.
Cheers,
Ashok.
Most of my tests were done under clipping level. I monitored it on a scope. The sound I was getting was like HF trash ( in the audible range). But since I put in the LPF I have not heard that yet. But I did make some other changes also.
Anyway it does sound nice if you don't push it too hard.
I've been toying with the idea of adding external components to improve on what is already there . Let me see how the 7294 fares.
Cheers,
Ashok.
Re: Re: Re: Some changes...
I already do a very high quality with superb PSSR regulation for the tube front-end of my commercial JLTi.
I am talking about something that I cannot quite go into detail here, but it will revolutionise tube audio right across the board (hush David).
Joe R.
carlosfm said:
Regulate.
I already do a very high quality with superb PSSR regulation for the tube front-end of my commercial JLTi.
I am talking about something that I cannot quite go into detail here, but it will revolutionise tube audio right across the board (hush David).
Joe R.
- Status
- Not open for further replies.